Feature Request: Basic Pre-filter on Metadata

Hi,

Really really love the work you've done here and have found RAGatouille so smooth to use.

I'm currently building RAG pipelines that can interact with text from company websites to ask basic questions, e.g. "Does this company do X?", "Does this company have an ESG policy?" etc. Since I want to ask the question about specific companies, I'm using an index per company, which works well.

I was wondering though if there was a way of pre-filtering the index on metadata or some basic syntactic query, before the semantic query.

i.e. Say I had segmented the text data from a company website based on page type, something like: ["about", "contact", "services"]. And had a pagetype variable in metadata. Is there a way to do a very basic filter on this e.g. pagetype == "about" / do you have an idea about best practices on how to do this?

Unsure if this is a strange use of RAGatouille, but I've been thinking about using for more complex / multi-hop queries, or just to get rid of data that I know won't be useful to my query - i.e. if I know my question is about what the company does -> "about", if I know I want an email -> "contact".

Thanks a lot! :)

No files, tests, or entry points are named in the issue. Start by locating RAGatouille's indexing and search APIs and how document metadata is stored; done should mean a caller can pre-filter by a metadata value before semantic retrieval.

Written by the indexing model from the issue text.